Agricultural Extension Education (Social Sciences & Management)

Option Format: 5 Options (5th Option Mandatory: 'अनुत्तरित प्रश्न')

## Instructions for AI Question Extractor & Taxonomy Mapper

When extracting questions and generating Schema v2.0 JSON packages, the AI model MUST strictly adhere to the following deterministic taxonomy protocol:

### 1. Deterministic Ground-Truth ID Assignment
Every extracted question MUST be mapped using the exact database integer IDs from this specification:
- `subject_id`: Integer ID of the Part / Section (e.g., `64` for General Hindi).
- `subject_name`: Canonical title of the Part (e.g., `सामान्य हिन्दी`).
- `subject_slug`: Canonical slug of the Part (e.g., `general-hindi`).
- `chapter_id`: Integer ID from the **Topic ID** column (e.g., `451`).
- `chapter_name`: Canonical title from the **Topic Name** column.
- `chapter_slug`: Slug from the **Topic Slug** column.
- `subtopic_id`: Integer ID from the **Subtopics (Level 3)** column if an atomic subtopic matches, otherwise `null`.
- `subtopic_name`: Subtopic title if mapped, otherwise `null`.
- `subtopic_slug`: Subtopic slug if mapped, otherwise `null`.
- `key_concept`: The specific atomic concept tested by the question (e.g., 'वृद्धि स्वर संधि', 'नागौरी गाय की विशेषताएं', 'अमोनियम सल्फेट में नत्रजन प्रतिशत').

### 2. JSON Mapping Target Schema
```json
{
"subject_id": 64,
"subject_name": "सामान्य हिन्दी",
"subject_slug": "general-hindi",
"chapter_id": 451,
"chapter_name": "संधि एवं संधि-विच्छेद",
"chapter_slug": "sandhi",
"subtopic_id": null,
"subtopic_name": null,
"subtopic_slug": null,
"key_concept": "वृद्धि स्वर संधि",
"question_type": "mcq_single"
}
```

### 3. Core Extraction Rules
- **Strict Section Boundary Isolation**: Respect examination paper sections. Map Hindi questions strictly to Subject ID 64, Rajasthan GK questions to Subject ID 34, Agronomy questions to Subject ID 74, Horticulture to Subject ID 75, and Animal Husbandry to Subject ID 65. Never cross-map questions across disciplines.
- **Strict Canonical Hierarchy**: Map every question extracted from papers for this examination strictly to one of the canonical Units/Chapters listed in the tables below. Do not fabricate or invent new chapter names or IDs.
- **Question Format Auto-Detection**: Identify and classify the question format strictly:
- Standard 4 or 5 options with 1 correct key: `mcq_single`
- Contains Statement I and Statement II or multiple statements: `statement`
- Contains List I and List II or matching columns: `match_list`
- Requires chronological arrangement or step sequence: `chronological_order`
- Contains Assertion (A) and Reason (R): `assertion_reason`
- **Semantic Keyword Anchors**: Use the 'AI Recognition Keywords' listed in the tables as primary semantic triggers. Matching question stems and option terminologies against these keywords provides high-confidence taxonomy assignment.
- **Out of Syllabus Preservation**: If an extracted historical PYQ belongs to a topic marked with `[OUT_OF_SYLLABUS]`, map it to that designated unit and flag `is_out_of_current = 1`. This preserves year-on-year historical practice.
- **Bilingual Terminology**: Ensure both Hindi (Devanagari) and English scientific terms are recognized according to the bilingual titles and keyword lists provided.